Structural damage identification method based on refined frequency band selection and energy intensity ratio

Abstract

Vibration-based energy indicators have been widely studied for structural damage identification, offering an advantage over traditional methods by avoiding reliance on modal parameters. However, their accuracy is highly sensitive to the selection of appropriate frequency bands. Conventional techniques typically require prior knowledge of damage states to determine the optimal frequency range, limiting their applicability in unsupervised damage identification. To overcome this limitation, this article proposes a structural damage identification method based on the energy intensity ratio of specific frequency band components. By analyzing the regularities of frequency band information changes caused by damage base on modal parameter perturbation theory, a damage information entropy function is constructed to accurately select sensitive bands without supervision. Using the optimally selected frequency bands, equivalent energy features related to acceleration, velocity, and displacement are constructed. The precise identification of damage is then achieved using the energy intensity ratio indicator. The feasibility, accuracy, and robustness of the proposed method are validated through numerical simulations of high-rise building and two typical experimental cases of frame structures. The comparative analysis and detection results indicate that, compared to traditional full-band energy indicators, the proposed method significantly improves the accuracy of damage localization and the sensitivity to damage degree changes, demonstrating promising prospects for engineering applications.